Water Extraction & Emissions

The proposed gas plants use the oldest, cheapest emissions control method - injecting water into the combustion chamber - requiring massive groundwater extraction. And even with that water, they emit over 8× the federal CO2 standard and qualify to operate only through a pollution exemption most Nova Scotians have never heard of.

Part 1: Water Extraction

175,000 L/hr
Peak water extraction per site
Marshdale EARD, Section 3.4.2
14–19
Production wells required at Marshdale
Marshdale EARD, Section 8.2
9–12
Production wells required at Salt Springs
Salt Springs EARD
50,000 L/hr
Wastewater discharge to surface water per site
Marshdale EARD, Section 3.4.2.4

How Much Water Are We Talking About?

Each 300 MW plant would extract up to 175,000 litres of groundwater per hour at peak operation - equivalent to roughly 770 gallons per minute. That water must be pumped from fractured bedrock aquifers through industrial production wells, processed through a demineralization plant, injected into the combustion chambers for NOx control, and then discharged as ~50,000 L/hr of wastewater to surface water.

The water infrastructure required at each site includes:

Source
Marshdale EARD (December 15, 2025), Sections 3.4.2, 3.4.2.4, and 8.2. The EARD describes 175 m³/hour (175,000 L/hr) as a "reasonable, conservative estimate" based on "combustion turbine model selection options available."

The estimated lifecycle cost of this water infrastructure across both sites is $36–76 million over the 20-year contract term, including $12–24 million in capital costs and $24–52 million in operating costs (demineralization chemicals, well pump electricity, effluent treatment, monitoring, and maintenance).

Source
Cost estimates based on SAMCO Technologies demineralization system pricing, Water Finder Canada well drilling costs for Nova Scotia, and comparable industrial water treatment installations. Full breakdown in water infrastructure cost analysis.

Impact on Cameron Brook

The proponent's own modelling shows the risk

The Marshdale EARD's groundwater modelling shows that 18–42% of pumping could come from depleting Cameron Brook. This is not an opposition claim - it is the project proponent's own analysis of what their proposed water extraction would do to a nearby watercourse.

The aquifer at Marshdale is fractured bedrock - a geology type rated medium-to-high risk for arsenic and manganese mobilization when subjected to sustained pumping. Large-scale groundwater extraction from fractured bedrock can change flow patterns, drawing contaminants from rock formations that are stable under natural conditions.

The EARD's stream depletion estimate of 18–42% represents a wide range of uncertainty. What is not uncertain is the direction: pumping 175,000 L/hr from this aquifer will affect nearby surface water. The only question is how much.

Source
Marshdale EARD (December 15, 2025), groundwater modelling results. Stream depletion range of 18–42% reported in the hydrogeological assessment sections.

Alternatives That Use Near-Zero Water

The proposed technology - SAC (Simple Annular Combustor) aeroderivative turbines with water injection - is not the only option. Two commercially proven alternatives dramatically reduce or eliminate water demand:

Technology NOx Control Method Process Water Use
SAC aeroCT (proposed) Water injection into combustor 175,000 L/hr per site
DLE aeroCT Premixed lean combustion (dry) ~0 L/hr
RICE engines Lean burn (closed-loop radiator cooling) <5 litres/hr
RICE with SCR Lean burn + urea injection for further NOx reduction ~500–2,000 L/hr

DLE (Dry Low Emissions) turbines use premixed combustion to achieve low NOx without any water injection. They are the same engine platform as the proposed SAC turbines - just with a different combustor module. GE Vernova's LM6000 PF DLE variant has accumulated over 11 million operating hours worldwide.

RICE (Reciprocating Internal Combustion Engines), such as the Wärtsilä 50SG, use lean-burn combustion in enclosed cylinders that produces approximately 90 ppm NOx without any water injection or aftertreatment. Cooling is closed-loop radiator-based, consuming less than 5 litres per hour. Even if SCR (Selective Catalytic Reduction) is added for stricter NOx limits, the water consumption is roughly 1% of SAC water injection volume.

Sources
GE Vernova: LM6000 PF DLE product specifications; "For DLE, water injection pumps aren't necessary." Wärtsilä 50SG spec sheet (May 2022): "NOx emission level 90ppm @15% O2 dry." Tehama County APCD Permit pto220 (California Power Holdings, LLC): 16 Wärtsilä engines with SCR, 9 ppm NOx limit, urea-water injection.

Both DLE and RICE are explicitly allowed by the IESO-NS RFP. The Functional Specifications (Exhibit T, Section 1) permit "combustion turbine generators or reciprocating engine generators." Neither alternative is a workaround - both are among the technology categories the procurement was designed for.

Source
IESO-NS Draft RFP, Exhibit T Functional Specifications (Hatch Engineering, Rev. 2, February 2026), Section 1.

Why Water Injection? Because It Is the Cheapest Option

Water injection into a SAC combustor is the oldest and cheapest method of controlling NOx emissions from gas turbines. It works by lowering the flame temperature, which reduces thermal NOx formation. The turbine hardware is simpler (no premixing fuel nozzles, no staged combustion), and refurbished SAC engines are widely available on the secondary market at lower cost than new-build DLE units.

The cost difference between a water-injection SAC configuration and a DLE configuration is estimated at 5–15% of the turbine package price - roughly $12–54 million across both sites (12 turbine units total). However, this premium is partially or fully offset by the $36–76 million in lifecycle water infrastructure costs that DLE eliminates.

Cost Category Water-Injection CT DLE CT Difference
Turbine hardware (both sites) $240–360M $252–414M +$12–54M for DLE
Water infrastructure capital $12–24M ~$1–2M -$11–22M with DLE
Water O&M (20 years) $24–52M ~$2–4M -$22–48M with DLE
Net lifecycle cost $276–436M $255–420M DLE saves $0–33M

In no scenario does the water-injection configuration save enough to justify the environmental impact of 14–19 production wells per site, 175,000 L/hr peak extraction from a fractured bedrock aquifer, and 50,000 L/hr effluent discharge to surface water.

Sources
DLE premium estimate: 5–15% based on general industry guidance for LM6000 SAC vs. DLE combustor options (exact pricing commercially confidential). Water infrastructure costs: order-of-magnitude estimates based on SAMCO Technologies demineralization system pricing, Water Finder Canada well costs, and comparable industrial installations. US EIA AEO2020/2025 capital cost data for turbine hardware ranges.

The Hydrogen Conflict: Why "Hydrogen-Ready" Locks In Water Injection

DLE combustion is capped at 35% hydrogen blend

DLE turbines achieve low NOx through carefully controlled premixed combustion. Hydrogen burns hotter and faster than natural gas, destabilizing the lean premix flame. Current DLE technology - including the GE LM6000 - is limited to approximately 35% hydrogen by volume. To burn 100% hydrogen, an engine needs a SAC combustor with water injection for NOx control.

The Tolling Agreement includes language about a "pathway to future operation without fossil fuels." The GE LM6000VELOX is rated for 100% hydrogen - but it uses a SAC combustor with water injection, not DLE. If the plants are to be "hydrogen-ready" for 100% hydrogen, the SAC + water injection configuration is locked in by the laws of combustion physics.

This creates a direct conflict:

There is an important counterpoint: NS Power's own IRP modelling explicitly rejected hydrogen-enabled fast-acting generation as uneconomic. The "pathway to clean fuels" language in the Tolling Agreement is aspirational with no enforcement mechanism, no timeline, and no funding commitment. The community is being asked to accept 20 years of groundwater extraction for a hydrogen future that the province's own utility has modelled and rejected.

Sources
GE Vernova: LM6000VELOX rated for 100% H2 with SAC combustor + water injection. DLE hydrogen blend limit: Black & Veatch, GE Vernova, and ETN (European Turbine Network) confirm DLE is capped at approximately 35% H2 by volume due to lean premix flame stability constraints. NS Power IRP Action Plan Update 2025: hydrogen-enabled generation rejected as uneconomic. IESO-NS Draft Tolling Agreement v1.1 (March 2026): aspirational "pathway" language with no binding commitment.

Part 2: The Pollution Exemption

8.2×
CO2 emission intensity vs. federal standard (on gas)
531 t/GWh vs. 65 t/GWh standard
12×
CO2 emission intensity vs. federal standard (on diesel)
777 t/GWh vs. 65 t/GWh standard
2× over
Annual emissions vs. federal cap at 25% capacity factor (gas)
349,000 t vs. 170,820 t cap
Until 2049
Exempt from any federal emission cap
CER SOR/2024-263, Section 3

How Dirty Are These Plants?

Simple-cycle gas turbines are the dirtiest form of gas-fired generation. They convert fuel to electricity at roughly 35% efficiency (the rest is waste heat), compared to 55–60% for combined-cycle plants. Their CO2 emission intensity on natural gas is approximately 531 tonnes per GWh - more than eight times the federal emission intensity standard of 65 t/GWh established by Canada's Clean Electricity Regulations.

On diesel backup fuel (which the RFP allows for up to 25% of operating hours), the emission intensity rises to approximately 777 tonnes per GWh - nearly 12 times the federal standard.

Fuel Emission Intensity vs. 65 t/GWh Federal Standard
Natural gas ~531 t CO2/GWh 8.2× dirtier
Diesel (#2 distillate) ~777 t CO2/GWh ~12× dirtier
Sources
CO2 emission factors: 40 CFR 98, Table C-1 (natural gas: 53.06 kg/MMBtu; distillate #2: 73.96 kg/MMBtu). Heat rates: EPA AP-42 Chapter 3.1 (simple-cycle gas turbine: ~10,000 BTU/kWh on gas, ~10,500 BTU/kWh on diesel). Federal emission intensity standard: Clean Electricity Regulations, SOR/2024-263, Section 9(1) - I_el = 65 t CO2/GWh for 2035–2049.

The "Planned Unit" Exemption

Canada's Clean Electricity Regulations (SOR/2024-263), published December 2024, cap CO2 emissions from electricity generation. Each generating unit receives an annual tonnage cap calculated from its nameplate capacity and the emission intensity parameter of 65 t CO2/GWh.

For a 300 MW plant, the annual emission cap would be:

300 MW × 65 t/GWh × 8,760 hours × 0.001 = 170,820 tonnes CO2/year

However, Section 3 of the regulation creates a special category called "planned units." If a project meets four milestones by December 31, 2025, and begins construction by December 31, 2027, it is completely exempt from the emission cap until December 31, 2049.

This is not a relaxed standard. It is no standard at all - for up to 25 years.

A "planned unit" faces zero federal emission limits from commissioning until the end of 2049. There is no reduced cap, no transitional limit, no phase-in. The exemption is binary: qualify, and you are completely unregulated. Miss the deadline, and you face an annual cap you cannot meet at any realistic operating level.

New Unit (no exemption) Planned Unit (with exemption)
Emission cap applies From 2035 Not until 2050
Years of unregulated operation 0 (from commissioning) ~20 years (2030–2050)
Can operate as peaker at 25% CF? No - 2× over cap Yes - no cap applies
Dispatch flexibility Constrained by tonnage ceiling Unlimited
Source
Clean Electricity Regulations, SOR/2024-263, Sections 3 (planned unit definition and milestones), 9 (emission limit formula), and 10 (planned unit exemption period). Full text: laws-lois.justice.gc.ca/eng/regulations/SOR-2024-263

The Capacity Factor Math: Double the Cap at Projected Use

The EARDs project these plants will operate at approximately 25% capacity factor. At that level, a single 300 MW plant on natural gas would emit approximately 349,000 tonnes of CO2 per year - more than double the 170,820-tonne federal annual cap.

Capacity Factor Fuel Annual CO2 (one plant) vs. 170,820 t Cap
100% Natural gas ~1,395,000 t 8.2× over
100% Diesel ~2,042,000 t 12.0× over
25% (projected) Natural gas ~349,000 t 2.0× over
25% (projected) 80% gas / 20% diesel ~389,000 t 2.3× over
15% Natural gas ~209,000 t 1.2× over
12% Natural gas ~167,000 t Just under cap

At any capacity factor above approximately 12% on natural gas (or ~8% on diesel), the plants exceed the federal cap. The regulation does allow Canadian offset credits, but they are capped at an additional 35 t/GWh (capacity-based), adding only ~92,000 tonnes to the allowance. Even with maximum offsets, a 300 MW plant at 25% CF on gas still exceeds the limit by approximately 86,000 tonnes.

These plants cannot operate at the levels the grid requires without the exemption.

Sources
Capacity factor projection: Marshdale EARD, Section 3.3. Emission calculations: 40 CFR 98 Table C-1 emission factors × EPA AP-42 heat rates. Annual cap formula: SOR/2024-263, Section 9(1) (C × I_el × 8760 × 0.001). Offset credit limit: SOR/2024-263, Section 28(2) (35 t/GWh for 2035–2049).

The Rushed Timeline Was Driven by the Exemption Deadline

To qualify as a "planned unit," a project must meet four milestones by December 31, 2025:

Milestone CER Deadline What Happened
EA information submitted to relevant authority Dec 31, 2025 EARDs registered Dec 22, 2025 - 9 days before deadline
Proponent owns or leases the site land Dec 31, 2025 IESO-NS acquired land options before any proponent was selected
Permit application info submitted Dec 31, 2025 Filed with EA registration
Equipment contracts ≥$10M signed Dec 31, 2025 No public information exists
Construction commenced (5th milestone) Dec 31, 2027 Pending
The $10 million question

The CER requires that contracts worth at least $10 million for equipment purchases be executed by December 31, 2025 - three months before the draft RFP was even issued (March 10, 2026). No contract has been disclosed. No equipment purchase has been announced. No line item appears in any public document. The Tolling Agreement's confidentiality provisions and the RFP's $20,000 data room paywall prevent verification.

If this milestone was not met, the planned unit qualification may be invalid - and the project's legal basis for operating without emission limits may not exist.

This deadline explains every aspect of the process that community members have questioned:

If proper planning and fieldwork had been done - if environmental assessments had taken the normal 12–18 months, if community consultation had been meaningful, if a competitive site selection had been conducted - the December 31, 2025 deadline would have been missed. And without the exemption, these plants cannot operate at the levels the grid needs.

Sources
Planned unit milestones: SOR/2024-263, Section 3. EA registration dates: NS Department of Environment, December 22, 2025. Procurement transfer: IESO-NS REOI (October 2025). Original IRP timeline: NS Power 10-Year System Outlook (July 2025).

The Contract Outlives the Exemption by 6–8 Years

The planned unit exemption expires December 31, 2049. After that date, the emission intensity parameter drops to 0 t CO2/GWh - the regulation's target is net-zero electricity.

But the Tolling Agreement runs for 25 years from the commercial operation date (~2030) to an expiry date of June 30, 2055. IESO-NS can unilaterally extend to year 26 (June 30, 2056) and request extension to year 27 (June 30, 2057).

Date What Happens
~2030 Plants commissioned, begin operating
2030–2049 No emission cap applies - planned unit exemption active
Dec 31, 2049 Planned unit exemption expires
2050 Emission cap drops to near-zero (offsets allow limited operation)
June 30, 2055 Tolling Agreement base term ends - 6 years after exemption expires
June 30, 2056–2057 With IESO extensions - 7–8 years past exemption

The plants go from no emission cap to a near-zero cap overnight. The Tolling Agreement does not address how the plants will operate - or whether they can operate - after 2050. Capacity payments are owed whether the plants run or not. Three possible outcomes after 2050, all bad for ratepayers:

  1. Plants cannot run - ratepayers pay capacity costs for idle assets for 6–8 years
  2. Plants buy massive carbon offsets - costs passed through to ratepayers on top of capacity payments
  3. Government weakens the regulation - the exemption becomes permanent
Sources
Tolling Agreement term: IESO-NS Draft Tolling Agreement v1.1 (March 2026), Sections 1.1 (definitions) and 9.1 (term and extension). Post-2049 emission parameter: SOR/2024-263, Section 10(3)(b) (emission intensity drops to 0 t/GWh). Post-2049 offset limit: SOR/2024-263, Section 28(2) (42 t/GWh for 2050+). Hydrogen pathway: NS Power IRP Action Plan Update 2025 rejected hydrogen-enabled fast-acting generation as uneconomic.

What 20 Years of Unregulated Operation Looks Like

At 25% capacity factor with an 80/20 gas/diesel fuel mix, both plants combined would produce annually:

~778,000 t
Annual CO2 (both plants, 25% CF, 80/20 gas/diesel)
Equivalent to ~169,000 cars on NS roads
~1,100 t
Annual NOx (uncontrolled)
Causes smog, respiratory illness, acid rain
~52 t
Annual PM2.5
Heart disease, lung cancer, asthma - no safe level

Over 20 years of unregulated operation, that is approximately 15.6 million tonnes of CO2 with no federal cap in place.

NOx from gas turbines is completely unregulated at the federal level

The Multi-Sector Air Pollutants Regulations (SOR/2016-151) regulate NOx from boilers and spark-ignition engines, but gas turbines operate on the Brayton cycle - they are explicitly outside the regulation's scope. The ECCC published guidelines in 2017 recommending NOx limits for gas turbines, but they are non-binding recommendations published under CEPA Section 54, not enforceable law.

On diesel backup fuel, the pollution profile is significantly worse: NOx is 2.7× higher than on gas, sulfur dioxide (SO2) - an acid rain precursor that is essentially zero on gas - is introduced, and particulate matter (PM2.5) is 2.7× higher. The RFP allows up to 25% of operating hours on diesel, but no enforceable regulatory condition caps diesel operating hours.

Sources
NOx, SO2, PM2.5 emission factors: EPA AP-42, Chapter 3.1 (Stationary Gas Turbines). CO2 emission factors: 40 CFR 98, Table C-1. Car equivalence: 4.6 t CO2/year per average Canadian passenger vehicle (NRCan). Federal NOx regulation gap: SOR/2016-151 (gas turbines excluded by definition); ECCC Guidelines for NOx from Combustion Turbines (November 2017, non-binding, CEPA Section 54). Diesel allowance: IESO-NS Draft Tolling Agreement Exhibit T, Functional Specifications v260220, Table 3-1.

The Bottom Line

The exemption is not incidental. It is the foundation the entire project stands on.

These plants cannot operate at their projected levels without the pollution exemption. And the entire rushed process - the pre-selected sites, the 9-days-before-deadline EA registrations, the undisclosed $10M equipment contracts, the compressed community consultation - was designed to qualify for that exemption before the window closed.

What They Say What's Actually Happening
"Fast-acting generation for reliability" Plants that cannot operate at projected levels without a pollution exemption
"No time to waste" The rush was to meet a Dec 31, 2025 exemption deadline - urgency the government created by transferring the procurement mid-stream
"Clean energy transition" 600 MW of fossil fuel generation exempt from emission caps for 20 years, emitting ~15.6 million tonnes of CO2
"Pathway to clean fuels" Non-binding language; NS Power's own modelling rejected hydrogen as uneconomic
"25-year contract" Contract outlives the exemption by 6–8 years with no plan for compliance
Primary Sources
Clean Electricity Regulations, SOR/2024-263 - Full text (Sections 3, 9, 10, 28). Marshdale and Salt Springs EARDs (registered December 22, 2025). IESO-NS Draft RFP v1.0 (March 10, 2026) and Draft Tolling Agreement v1.1. NS Power IRP Action Plan Update 2025. EPA AP-42, Chapter 3.1; 40 CFR 98, Table C-1. Multi-Sector Air Pollutants Regulations, SOR/2016-151. ECCC Guidelines for NOx from Combustion Turbines (November 2017).